Capital Flows as an Offset to Monetary Policy: The German Experience

THE FLOTATION of the deutsche mark in May 1971 was preceded by a period in which the authorities of the Federal Republic of Germany attempted vigorously to pursue an independent monetary policy. In particular, German interest rates were kept well above Euro-dollar levels, with the result that there were substantial inflows of capital. The Bundesbank tried to neutralize the accompanying rise in bank reserves, largely by means of periodic increases in reserve requirements. In retrospect, it is evident that this neutralization policy was unsuccessful because it was offset by the induced accumulation of large foreign exchange reserves.1


THE FLOTATION of the deutsche mark in May 1971 was preceded by a period in which the authorities of the Federal Republic of Germany attempted vigorously to pursue an independent monetary policy. In particular, German interest rates were kept well above Euro-dollar levels, with the result that there were substantial inflows of capital. The Bundesbank tried to neutralize the accompanying rise in bank reserves, largely by means of periodic increases in reserve requirements. In retrospect, it is evident that this neutralization policy was unsuccessful because it was offset by the induced accumulation of large foreign exchange reserves.1

THE FLOTATION of the deutsche mark in May 1971 was preceded by a period in which the authorities of the Federal Republic of Germany attempted vigorously to pursue an independent monetary policy. In particular, German interest rates were kept well above Euro-dollar levels, with the result that there were substantial inflows of capital. The Bundesbank tried to neutralize the accompanying rise in bank reserves, largely by means of periodic increases in reserve requirements. In retrospect, it is evident that this neutralization policy was unsuccessful because it was offset by the induced accumulation of large foreign exchange reserves.1

This paper examines the effects of monetary policy in Germany on capital flows during the period 1963–70. It reports tests of the use of changes in reserve requirements to neutralize the monetary effects of capital inflows. An important finding of the study reported here is that changes in the average reserve requirements—the major control in German monetary policy—are substantially and rapidly offset in their effect on bank liquidity by capital inflows recorded mainly in the errors and omissions component of the balance of payments.2 This suggests that the monetary policy options available to a country such as Germany, which allows free movement of financial capital across its borders, are somewhat limited under fixed exchange rates.

Framework of the study

In an economy such as Germany’s, which has a high degree of international capital mobility, the demand for money may be satisfied through either domestic or foreign sources. Restrictive monetary policies—such as independently high interest rates, increases in required reserve ratios, and special reserve requirements on the growth of foreign liabilities—may bring about little reduction in the domestic money supply but may rather cause domestic money demand to be satisfied indirectly through foreign sources. This may occur in one of two ways: (1) German banks may borrow abroad, or (2) if the banks are effectively controlled, corporations and individuals may borrow directly from foreign banks and corporations and then place the funds in German banks.

Most of the recent studies of capital flows 3 have focused on changes in international interest rate differentials as the basis for short-term capital flows. However, it is believed that this approach needs modification to reflect current conditions of capital market integration in parts of Europe and North America. As mentioned above, official monetary restrictions tend to prevent German banks from increasing loans in accordance with demand and may simply lead the frustrated borrower to turn directly to foreign banks or to a foreign affiliate, with the result that a capital inflow (either recorded or unrecorded) will occur without any change in the observed German interest rate. Thus, if German and Euro-dollar rates are initially in equilibrium (in the sense that all portfolios have the desired mix of foreign and domestic assets and liabilities, and the projected growth in these assets and liabilities is consistent with actual or expected yields), the termination of an expected German loan in view of the increased cash reserve requirements of German banks may lead the customer to turn to foreign sources for his monetary needs. Since Euro-banks feel no impact of the German reserve requirement policy, a rational borrower may turn immediately to them at given interest rates.

This point may be illustrated by a hypothetical case. If a group of banks in Upstate New York were suddenly singled out for reserve requirement increases, money would flow into Upstate New York, not because interest rates rose in response to the shortage of funds but because frustrated Upstate borrowers knew that they could turn to outside banks. Thus, the capital inflows to Upstate New York would be explained by credit restrictions and not by observed interest rate differentials. To state this another way, a very fast adjustment to any excess demand for money would assure that changes in interest rates in Upstate New York would not be observed in response to the monetary restrictions. It can be argued, therefore, that observed interest rate differentials become less important in explaining capital flows as the integration of the domestic capital market with the foreign capital market becomes greater. On the other hand, credit restrictions on domestic banks become more important in explaining capital inflows as capital market integration increases, but at the same time they become less efficient as a monetary instrument. An alternative method for explaining the use of the change in reserve requirements as an independent variable in capital flow equations is that it is a good proxy for short-term credit rationing (i.e., rationing which is avoided by borrowing abroad).

It is difficult to analyze the errors and omissions component of the balance of payments (called “unclassifiable items” by the Bundesbank and used thus in this paper) because, by definition, little is known directly as to the nature of these transactions. However, there is evidence that errors and omissions entries account for foreign borrowing and lending by corporations, including leads and lags in export and import payments. In times of speculation, such errors and omissions can be very substantial. For example, in 1970 errors and omissions amounted to 44 per cent of the $6.2 billion increase in net gold and foreign exchange holdings of the Bundesbank, with recorded net short-term capital flows accounting for 62 per cent and a debit in the basic balance accounting for minus 6 per cent. Thus it is important to be able to test the sensitivity of the errors and omissions component to changes in interest rates and in other conditions of monetary policy in Germany.

Changes in required reserves of German banks have frequently been intended to neutralize the bank liquidity resulting from the accumulation of foreign currency holdings. While bank reserves can be reduced on any given day by an amount specified by the central bank, such a reduction need not have a sustained effect on bank liquidity. As mentioned briefly above, the important factor is whether the foreign lending absorbs any demand for loans that is frustrated by the credit restrictions associated with the increase in required reserves. The important question, then, is to what extent neutralization policies are offset. This matter will be dealt with in depth below.

The econometric approach taken here is to use ordinary least squares estimates of the capital flow equations, with changes in the covered differential and reserve requirements as independent variables. This approach is subject to simultaneous equations bias, because German interest rates and forward exchange premiums (and possibly the Euro-dollar rate as well) may be affected by capital flows. The exact nature of the bias is made clear in a later section. The analysis indicates, however, that the results underestimate the extent to which monetary policy is offset by capital flows.

Impact of exogenous disturbances in open economies

It is generally agreed that under conditions of high international capital mobility and fixed exchange rates, monetary policy is relatively ineffective as an instrument for regulating fluctuations in domestic economic activity.4 For a better understanding of this argument, it will be useful to discuss briefly the consequences of fluctuations in external and internal monetary conditions and the options open to policymakers wishing to offset these disturbances.

Fluctuations in external interest rates

Suppose that the German economy initially had its level of domestic activity and interest rates at satisfactory values, and that there was a fixed exchange rate without exchange controls. Suppose further that there was then a reduction in Euro-dollar rates, so that Germany received massive inflows of funds. Bank controls, such as increases in reserve requirements, reserve ratios on the growth of foreign liabilities, and official swap rates inducing banks to lend funds abroad, would not necessarily prevent funds from flowing in but would affect the composition of the inflow. As a result, foreign banks and nonbank enterprises would be recorded as the main source of the inflow. In this case, if capital inflows are to be prevented, either the German interest rate must fall in line with the Euro-dollar rate, which would conflict with domestic objectives, or the forward discount on the deutsche mark would have to rise in such a way that the covered differential in favor of Germany would not persist. However, if speculators have the view that upward movements of the deutsche mark are a real possibility, then it may not be feasible to push the deutsche mark to a forward discount without inducing considerable purchases of forward deutsche mark. On the other hand, when the deutsche mark is strong and it is desirable to have lower interest rates in Germany than abroad, it will be feasible to remove the covered differential by allowing the forward deutsche mark to sell at a premium, thereby preventing outflows that would otherwise result from the lower domestic interest rates.5

Fluctuations in money demand

Suppose that the demand for real money balances varied over time. One possibility is that the domestic price level would adjust so that the desired level of real balances would be obtained. However, given a short period of analysis (the month), it is reasonable to assume that the adjustment process involves interest rates and nominal monetary stocks, since the commodity market is likely to adjust much more slowly than the international capital market. In this case, capital inflows will accommodate the increased demand for money, unless the domestic assets of the central bank are increased to satisfy the rise in money demand.6

Looking back on German monetary behavior, it appears that the growth in money demand has been largely satisfied through foreign sources.7 This external means of augmenting the money supply has the inherent difficulty that the level of foreign assets that is consistent with domestic monetary equilibrium may be inconsistent with the exchange rate equilibrium. If the central bank wishes to prevent the accumulation of foreign reserves, it will have to increase its domestic assets so that the increased demand for money may be satisfied from domestic sources. Thus, under conditions of high capital mobility, the authorities are left with two options, neither of which they may find attractive. In Germany in 1970 the authorities chose not to increase domestic assets to satisfy the increase in money demand, with the result that the foreign assets of the central bank increased substantially.

Econometric estimates of capital flows

The portfolio approach to capital flows 8 argues that the potential participants in the German capital market will allocate their stock of wealth, at time t, between German and foreign securities according to the expected net yields and variances in the markets. Accordingly, for a given stock of wealth, changes in expected yields will cause capital flows into or out of Germany. On this assumption, a change in the covered yield differential between Germany and the Euro-dollar market may cause a change in expected yields, and this in turn will cause a once-for-all capital flow with the result that, by the end of the period, portfolios will contain German and foreign securities in the required proportions and there will be no further flows resulting from the yield differentials.

This same approach argues that if demand functions are homogeneous of degree one with respect to wealth, then changes in the stock of wealth, with yields in Germany and other countries being unchanged, will be allocated between German and foreign securities in the proportions previously maintained. In this instance, capital flows will result from the divergent levels of interest rates in the two markets. The absence of data on the identity and wealth of participants in the market prevents the computation of a series on Wt and accordingly, the results will suffer from mis-specification. The approach taken in this study is to omit the wealth variable from the equations and to use the changes in covered interest rate differentials—but not the levels of the differentials—as independent variables. In other words, a modified stock adjustment model of capital flows is tested here, the modification being the inclusion of the reserve requirements variable and the omission of the wealth variable.9

The change in the covered interest rate differential (current and lagged) is used as an independent variable. A difficulty in using this approach is that it is incomplete in times of speculation, since in order to capture the speculative flows it would be necessary to use as independent variables the uncovered interest rate differentials, the forward premium, and a separate variable used as a proxy for the expected change in the exchange rate.10 An additional difficulty is that the single equation estimates are subject to simultaneous equations bias, although it is shown below that the conclusions of the paper would be strengthened by removing the source of the bias.

A key role in the following regressions is played by the monetary instrument ΔRR*, the exogenous component of the total change in required reserves ARR. Required reserves may vary because of changes either in the required reserve ratio or in the stock of liabilities against which reserves must be held; i.e.,


where rt is the average required ratio of reserves to total liabilities (L) and where Lt is the average liabilities outstanding in month t. The reserve ratio is always changed effective the beginning of the month in Germany; hence, ΔRRt is the change in average reserves required in month t.11 The component of change in required reserves which may be regarded as an instrument is ΔRRt*=(rtrt1)Lt, since it accounts for the change in required reserves caused by changes in the reserve ratio.12 We regard rt − rt-1 as being zero in those months in which there was no change in the required ratios—even though the reported average ratios change slightly because of shifts in deposits between classes of banks subject to different reserve ratios. Since ΔRRt is in the same dimension as capital flows (millions of deutsche mark per month), the instrument is expressed in these terms.

Simultaneous equations bias13

A difficulty with using the change in the covered interest differential as an independent variable in the capital flow equations is that there is a presumption that the covered differential will itself be affected by the flows. In particular, both the spot and forward exchange rates are likely to be sensitive to capital flows. In what follows it is assumed that the covered differential is a function both of capital flows and of exogenous factors subsumed in the variable x.

Suppose that the true system is represented by the equations (1) and (2):


where STC is short-term capital inflows; iG is the German three-month interbank rate; iEuro is the three-month Euro-dollar rate; f is the forward premium on the deutsche mark; ΔRR* is the change in required reserves; and x is a proxy for other influences on the forward premium.

If the usual assumptions are made regarding the error terms, namely:




then it can be shown that the ordinary least squares estimate of β2(β^2) is always biased downwards and that the ordinary least squares estimate of β1(β^1) is also biased downwards, provided γ1, <0. The probability limits of the biases are given below:


where M¯xx is the (finite) asymptotic limit of the moment Mxx.

In other words, as long as short-term capital inflows reduce the covered differential and thereby serve as a stabilizing influence, the estimates of β1 and β22 are less than the true values. Ideally, it would be appropriate to estimate both equations (1) and (2) by two-stage least squares or other methods that are free of simultaneous equations bias. However, this would require information on x, where x is interpreted as a vector of all exogenous variables influencing the German interest rate, the Euro-dollar rate, and the forward premium or discount.

It should be noted that the sign of the bias is independent of the x factor, since γ2 only appears in the denominator of the above expressions. The size of the bias is dependent on the x factor. Finally, if capital flows are destabilizing in the sense that γ> 0, for example, then if capital inflow induces a forward premium, it follows that β^1 (but not β^2) is biased upwards if γ1β1 < 1

To the extent that the x factor in the omitted equation (2) represents variables explaining speculative flows, it follows that the total explanatory power of the equation as measured by the R¯2 coefficient will be substantially reduced. These flows can be in massive amounts, and omission of the x factor means that there is a substantial unexplained variance. However, the objective of this study is to obtain useful estimates of the offset to monetary policy caused by capital flows; hence, it is mainly concerned with the significance of the estimated coefficients and the sign of the bias. It is noteworthy that Branson and Hill in a recent study of German capital flows 14 omitted the speculative period (from 1969 on) and thereby obtained substantially higher R¯2. However, the interest rate coefficients obtained by Branson and Hill are less significant than those in this study.


The period studied is from January 1963 to January 1971, using monthly data. Since visual analysis suggested that purely seasonal factors caused capital flows in the months of December and January, seasonal dummies were used for these months.15 The coefficients were not substantially affected by allowing for these elements. The results for the equations using unlagged interest rate data are given in Tables 1, 2, and 3. In Table 2 the results of adding a dummy for the speculation in May 1969 are given. Table 3 shows the additional impact of adding a variable representing the divergence between the official forward rate made available to banks and the market forward rate. In this case, a test was made as to whether this form of interference in the forward market was effective in inducing additional net capital outflows.16 Trade flows were not used as independent variables. Branson and Hill17 did include trade variables and found them to be relatively insignificant.

Table 1.

Short-Term Capital Flows (STC) and Unclassifiable Items (UNCL) in the Balance of Payments as Explained by Changes in Covered Yield Differentials, Reserve Requirement Changes at Beginning of Month, and Seasonal Factors, January 1963–November 1970 1

article image

The figures in parentheses are t-coefficients. DW is the Durbin-Watson statistic; iG = German three-month money market rate; iEuro = three-month Euro-dollar rate; f = market swap rate (forward exchange premium on the deutsche mark); ΔRR* = the change in required reserves due to changes in the average reserve ratio. With the exception of the coefficient on ΔRR*, the units of the coefficients are millions of deutsche mark. The coefficient on ΔRR* measures the extent to which the change in reserve requirement was estimated to be offset within the month; for example, a coefficient of .8193 indicates an 82 per cent offset.

Table 2.

STC and UNCL in the Balance of Payments as Explained by Changes in Covered Yield Differentials, Reserve Requirement Changes at Beginning of Month, and Seasonal Factors, January 1963–November 1970, with Addition of Dummy for May 1969 1

article image

In May 1969 net inflows of STC plus UNCL amounted to DM 10.3 billion, reflecting speculative inflows that subsided with the government announcement that there would be no revaluation. See footnote on Table 1 for explanation of symbols and units used.

Table 3.

STC and UNCL in the Balance of Payments as Explained by Changes in Covered Yield Differentials, Reserve Requirement Changes at Beginning of Month, and Seasonal Factors, January 1963–November 1970, with Addition of Variable for Swap Policy1

article image

Swap = market swap rate minus official swap rate offered to banks by the Bundesbank. Positive swap rate is an inducement to banks to place funds abroad. See footnote on Table 1 for explanation of symbols and units used.

Interest differentials

A noticeable feature of the results is that the coefficients for the changes in the covered differential are significant at better than the 1 per cent level for both short-term capital flows (STC) and the unclassifiable items in the balance of payments (UNCL), with the interest rate changes being more significant for STC than for UNCL. Taking these results at face value, a 1 per cent covered differential in favor of Germany produced a total capital inflow of approximately DM 1 billion in the same month. The equation does not allow for lagged adjustment of capital flows to changes in yield differentials, yet the lagged responses are of a substantial magnitude, as is shown later. Addition of a dummy for the speculative flows of May 1969 (see Table 2) reduces the coefficients to a value that indicates a capital inflow of DM 770 million resulting from a 1 per cent change in the covered differential. The coefficient falls to DM 749 million when a separate variable for “swaps” policy is included (see Table 3).

Reserve requirements

The regressions in Table 1 indicate that a rise in the required reserve ratio at the beginning of the month which is sufficient to cause a DM 1 billion increase in required reserves is followed by capital inflows that same month of approximately DM 820 million. The majority of this induced inflow (DM 655 million) is seen as coming through the unclassifiable items in the balance of payments. The size of this offset is reduced, once a broader specification of the equation is taken. For example, from Tables 2 and 3 it can be seen that the value of the coefficient falls from .820 to .614 and .617, respectively. It should be recalled at this point that simultaneous equations bias is present in these equations and that these estimates understate the extent to which the effect of reserve requirement changes was offset.

Other variables

The swap coefficient in Table 3 measures any additional effect on capital flows caused by a difference between the official and market forward rate. A 1 per cent differential between the two rates (in favor of dollars) proved sufficient to induce monthly outflows of DM 958 million, the standard error being DM 490 million. For example, by offering to sell the banks forward deutsche mark at a discount relative to the market rate, the Bundesbank appears to have been able to generate additional capital outflows.

The coefficients on the seasonal dummies indicate that in December there are substantial outflows through the unclassifiable items (errors and omissions) account, with these outflows being partially offset by recorded capital inflows. In January the recorded short-term capital outflows are found to be significant, although partially offset by inflows recorded in the errors and omissions component of the balance of payments. It should be noted that the constant term is significant and positive in all the equations (lagged and unlagged), presumably reflecting the (omitted) increase in wealth over time and the net positive wealth elasticity of demand for German securities. In terms of the R¯2 statistics, it should be noted that the equations for UNCL do better than the equations for STC, reflecting the high level of significance of the reserve requirement coefficient. Serial correlation, as measured by the Durbin-Watson statistic, is not evident to any serious extent in the equations.

The model of lagged adjustment

The approach adopted here is to allow a lagged pattern in the adjustment of portfolios to changes in the covered differentials. This approach is justified if, for various reasons, expected yields are a function of lagged values as well as current interest rate conditions. The path of adjustment is permitted to assume the form of an unconstrained second-order polynomial, i.e., the coefficients must lie on portions of a parabola. The equations were estimated using three-month and six-month distributed lags.18 An additional variation was to partition the change in the covered differential into its three components: the change in the Euro-dollar rate, the change in the German three-month money market rate, and the change in the forward exchange premium. The reserve requirement variable was left in its unlagged form, on the ground that the inclusion of the variable was justified because reserve requirement increases, for example, caused temporary domestic disequilibrium (or rationing) which rapidly diverted borrowers to foreign capital markets.

The results for the three-month lag specification (Table 4) indicate that the adjustment process may have been complete after three months. However, this suggestion was not confirmed by the results for the six-month regression (Table 5). A comparison of the coefficients on the total effect over three months and six months shows that the effect after six months is substantially greater than after three months, with the flows recorded in the errors and omissions section being more significant after lags of three, four, and five months. The most important result is that the coefficient measuring the extent to which reserve requirements are offset is not significantly changed (in magnitude or significance) by changes in the specification of the equations. The total percentage offset within the month is revealed as 85.8 per cent and 69.5 per cent in the six-month and three-month equations, respectively. Again, it is noted that the effect occurs mainly through the errors and omissions component of the balance of payments.

Table 4.

STC and UNCL in the Balance of Payments as Explained by Use of a Three-Month Distributed Lag, January 1963–January 1971 1

article image

The results in Tables 4, 5, and 6 included the two added observations that were available at the time these latter equations were estimated. The swap variable was found to be only weakly significant when incorporated in the equations with distributed lags. The other estimates were not affected. See footnote on Table 1 for explanation of symbols and units used.

Table 5.

STC and UNCL in the Balance of Payments as Explained by Use of a Six-Month Distribution Lag, January 1963—January 1971 1

article image

See footnote on Table 1 for explanation of symbols and units used.

Several factors suggest that it is desirable to separate the components of the covered differential. First, the dominant changes in the covered differential have come from the Euro-dollar rate, with the German interest rate and forward premium showing less variation. Second, there is reason to believe that the German three-month interbank money market rate is not always the best indicator of market conditions in the nonbank sector. Thus, it is possible that the changes in covered differential may hide information that would be revealed in a partitioned differential. An additional factor is that Euro-dollar rates may be more independent of capital flows than the other components of the covered differential. An argument against partitioning the covered differentials is the possible collinearity of the components, but this does not appear to be a problem in this case.

The results for the equation explaining recorded short-term capital flows in terms of the lagged partitioned covered differential are given in Table 6. The equation for errors and omissions is not reported, because it revealed very weak results for the lagged interest rate terms although the reserve requirements coefficient was unaffected by the partitioning. The results of the reported equation indicate that changes in the Euro-dollar rate are the dominant element in explaining recorded short-term capital flows. The coefficients on the changes in the German interest rate and the forward premium have the expected signs and are significant, except that the size of these two coefficients is less than for the Euro-dollar rate. This result probably reflects the simultaneous equations bias discussed earlier, in that capital flows tend to affect interest rates and forward and spot exchange rates in Germany but have relatively little effect on Euro-dollar rates. Therefore, the Eurodollar coefficients are probably the best estimates of the true interest rate sensitivity of capital flows. The total inflows resulting from a once-and-for-all 1 per cent fall in the Euro-dollar rate (with other rates constant) are found to add to DM 5.3 billion after six months, an amount considerably higher than that revealed in previous estimates. The pattern of lagged response is that 25 per cent of the effect is recorded concurrently, with 20 per cent, 15 per cent, 12 per cent, 10 per cent, 9 per cent, and 9 per cent being recorded in succession over the next six months. To state this another way, 61 per cent of the adjustment to a change in Euro-dollar rates is completed with a two-month lag, 73 per cent with a three-month lag, etc. The statistical significance of the coefficients is found to diminish as the length of the lag increases.

Table 6.

Short-Term Capital Flows as Explained by Partitioned Covered Differential (Lagged), January 1963–January 1971 1

article image

See footnote on Table 1 for explanation of symbols and units used.

The pattern of lag coefficients on the German interest rate variable is very similar to that for the Euro-dollar rate; the pattern on the forward premium, however, indicates a more rapid speed of adjustment, with 54 per cent occurring within a one-month lag.

Table 7 uses the results of Table 6 to simulate the effect on capital inflows of a succession of three monthly reductions of 1 per cent in the Euro-dollar rate, while the German rate and the forward premium remain unchanged. The cumulative effect after nine months is found to create inflows to Germany of DM 16 billion.

Table 7.

Simulation of the Effect on Capital Flows to Germany of Three Successive Reductions in the Euro-Dollar Rate, with German Interest Rate and Forward Premium Held Constant

article image

An alternative view of German monetary policy

In a recent study Manfred Willms 19 argues that German monetary policy has been successful in substantially neutralizing the effects on base money of changes in the foreign assets of the Bundesbank. However, it appears that a critical causal connection has been improperly reversed in the study.

The basic finding is summarized in his equation:


We use Willms’ terminology, i.e.,


where B = total monetary base; FR = net foreign assets of Bundesbank; B8 = monetary source base; Br = adjustment component for changes in required reserves; r = reserve ratio (averaged over all banks); and RV = deposit liabilities of banks.

The left-hand side of the basic equation is viewed as the portion of base money that is controlled by the authorities (i.e., total change in base money minus the change due to foreign asset accumulation). The right-hand side is the change in base money due to changes in foreign assets. The hypothesis advanced by Willms is that the controlled component has been so varied as to offset about 86 per cent of the changes in base money due to changes in foreign assets.

The difficulty with the Willms study is that the results are consistent with the totally different interpretation advanced in this paper—namely, that the changes in required reserves induced the capital inflow. Because the data used by Willms are quarterly, there is no presumption that the change in foreign assets precedes the officially induced changes in base money, or vice versa. In order to test properly the connection between changes in base money due to reserve requirement changes and changes due to accumulation of foreign assets of the Bundesbank, it is necessary to use monthly data in the manner developed in the earlier sections of this paper. If increases in reserve requirements, always implemented at the beginning of the month, tend to be followed by capital inflows equal to about 80 per cent of the increase in reserve requirement, as is found to be the case, then the inference is that it is the capital inflow that offsets the changes in reserve requirements and not the reverse, as Willms would have it.

Summary and conclusions

The main finding of this study is that the impact on base money of changes in average reserve requirements tends to be offset by capital flows amounting to about 80 per cent20 of the change in required reserves. In Germany, where the increase in required reserves is always implemented on the first day of the month, this study indicates that the 80 per cent offset takes place by the end of the month. The most reasonable interpretation is that corporations turn rapidly to foreign sources if they anticipate that loans will be refused; the subsequent capital inflow is unrecorded and appears as an unclassifiable item in the balance of payments (i.e., errors and omissions). This behavior was particularly noticeable in 1970 when the substantial reserve requirement increases were being rapidly offset by inflows (see Appendix). It now appears that the Bundesbank has come to essentially the same conclusion, as is made clear in the June 1971 issue of the Monthly Report of the Deutsche Bundesbank, page 7. Referring to the stabilization possibilities following the flotation of the deutsche mark, the Bundesbank states: “The Bank is now released from the compulsion of having to create central bank money by the purchase of foreign exchange…. The Bundesbank thus no longer has to fear that its restrictive course in credit policy is more or less automatically undercut by money inflows from foreign countries.”

Despite the finding regarding the offset to acts of monetary restriction, this study suggests that monetary policy has some impact, since some people and institutions continue to borrow at the penal domestic rates, most likely small borrowers with limited access to foreign markets, e.g., borrowers in the housing and construction sector. However, it is clear that large entrepreneurial borrowers have little difficulty in avoiding domestic credit restriction and thus it is to be expected that monetary stringency will have quite uneven impacts.21

There is evidence of considerable sensitivity of recorded short-term capital flows to changes in interest rate differentials, with the coefficients on the change in the Euro-dollar rate being of a substantial size. In addition, there is evidence of some lag in adjustment of portfolios to changed interest rate conditions, although about 75 per cent of the effect occurs within three months.

The total explanatory power of the regressions estimated in this study is substantially less than would be obtained by supplementing the analysis with variables capturing speculative capital flows, but the explanation of speculative flows is likely to be a complicated and possibly somewhat arbitrary procedure. These unexplained flows, which account for a large part of the variation in the time series (in particular the flows of 1969), have caused the R2 statistics to be substantially less than they would have been otherwise. However, the finding in regard to reserve requirements does help to explain some of the factors leading up to speculative crises. The attempts to neutralize capital flows resulting from pursuit of independent monetary policies are found to be substantially offset by the capital flows, with the result that a relatively small degree of monetary independence is obtained at the expense of a large amount of fluctuation in the foreign reserves position. Consequently, the attempts to achieve internal balance appear to have aggravated the external imbalance situation.


German Monetary Policy in 1970–71

The events prior to March 1970 in the Federal Republic of Germany largely reflected adjustments associated with the outflow of capital after the 1969 currency revaluation. Therefore, this brief survey of German monetary policy focuses on events since March 1, 1970.

On March 6, 1970 restrictive monetary policy measures were adopted by the Bundesbank, the main features being a rise in the discount rate from 6 per cent to 7.5 per cent and a supplementary marginal reserve ratio of 30 per cent on bank external liabilities.22 These measures were directed against domestic inflationary trends, as reflected, for example, in a 6.2 per cent rise in producer prices of industrial products over the 12 months to February 1970, the corresponding figure for the preceding year being 0.3 per cent. The discount rate increase was interpreted by the Bundesbank as “confirmative” of the increases in market interest rates due to reduction in bank liquidity following the exodus of speculative funds left over from 1969—a reduction which had the passive support of the Bundesbank.23 It is indeed true that interest rates in Germany had been increasing over the March period, despite the declining Euro-dollar rates. To understand this it is helpful to look at factors affecting domestic money supply and demand.

The upward pressure on interest rates in Germany prior to the March 1970 increase in the discount rate can probably be largely attributed to a sharp rise in the domestic demand for money and the outflow of speculative funds following the currency revaluation. Wages and salaries, for example, showed strong increases over this period, as is reflected in Table 8. These data are suggestive of a strongly rising demand for money. On the supply side there was a striking fall in unused rediscount quotas of DM 6.7 billion over the four-month post-revaluation period to the end of February 1970 (cf. increases of DM 0.2 billion for the corresponding period in the preceding year). The increase in the demand for money was satisfied to a large extent by domestic credit institutions. In addition to using their rediscount quotas, these institutions also drew to a substantial extent from foreign sources. The short-term position of German domestic banks in relation to foreign countries changed from net foreign assets of DM 3.8 billion at the end of November 1969 to net foreign liabilities of DM 4.8 billion at the end of February 1970. The bulk of this change represented increased liabilities to foreign banks.24 The lending by the private banking system to the nonbank sector rose by DM 15 billion over the four months prior to February 1970, an increase in loans which is highly consistent with the postulated sharp increase in the demand for money over this period.

Table 8.

Percentage Increase in Gross Wages and Salaries in Germany over Corresponding Period of Preceding Year

article image
Source: Deutsche Bundesbank, Monthly Reports.

It is clear that the period prior to the March increase in the discount rate was characterized by a sharp increase in the demand for money. This demand had been satisfied by the banking system’s borrowing both from the Bundesbank (using rediscount quotas) and from foreign banks by approximately equal amounts. Domestic enterprises appeared to satisfy their demand for money from domestic banks; in fact, over this November to February period there was a net repayment of DM 2.91 billion debts to foreign lenders by domestic enterprises.

The main objective of the officially induced rise in German interest rates in March 1970 was to reduce private expenditures by making it more expensive to borrow. This Keynesian approach faced major obstacles. In the first place there are no exchange restrictions in Germany, and banks and enterprises are free to borrow abroad if it is cheaper. Second, even if domestic banks are induced not to use foreign credit (as by the marginal reserve requirement of 30 per cent introduced April 1970), this may merely have the effect of encouraging domestic enterprises to borrow directly from foreign sources. Third, foreigners are likely to place funds with German banks if covered German interest rates exceed external yields. Thus, although it is possible for the German authorities to raise interest rates independently, there is no assurance that this would exert a dampening effect on the domestic economy, although there is an assumption that it would lead to short-term capital inflow.

Events following the rise in discount rate, March 1970

What actually happened was that the rise in German interest rates, triggered by the rise in the discount rate, caused increased borrowing abroad by banks and enterprises, with the latter taking a large part of the credit. This phenomenon was no doubt due to the 30 per cent marginal reserve requirement on foreign liabilities, which made it more sensible to borrow directly from the foreign banks rather than indirectly from domestic banks.25

From the beginning of March to the end of June 1970 (the month prior to the restrictive monetary policy action of July 1) recorded short-term capital inflow amounted to DM 4.2 billion, to which should be added the sum of the unclassifiable transactions (best interpreted as short-term capital inflows) of DM 2.8 billion. Thus, over the four months prior to the end of June 1970 there was a DM 7.0 billion total inflow of foreign money, of which only DM 1.2 billion came through German banks. The net effect was that German money demand was satisfied through foreign sources and was not frustrated in the manner intended by the German authorities. Bank lending to nonbanks rose DM 13.4 billion over the four-month period prior to the end of June 1970, and there were few signs that the monetary policy had reduced the level of aggregate demand. The overall effects of the measures announced on March 6, 1970 appear to have been substantial capital inflow and diversion of a large part of the borrowing by German enterprises to channels other than the German banking system.

Increases in reserve requirements, July 1970

The policy measure of increasing reserve requirements can be interpreted as an attempt to neutralize previous increases of the domestic banking system in foreign liabilities,26 the intention being to absorb about DM 3 billion of the increased bank liquidity.27 It should be noted that this increase in bank liquidity was largely the result of the monetary policy enacted in March 1970 to maintain yield differentials in favor of Germany, thereby attracting foreign money into the German banking system.

The other aspects of the tight monetary policy were continued, although on July 16 the discount rate was reduced from 7.5 per cent to 7 per cent and other official interest rates were reduced by 0.5 per cent. However, this reduction was less than was required to bring German interest rates in line with Euro-dollar rates. For example, over the two-week period during which the reduction in the discount rate took place, the call money rate in the Euro-dollar market fell from 8.75 per cent (week ending July 10) to 7.09 per cent (week ending July 24) with the one-month and three-month Euro-dollar rates falling by 0.83 per cent and 0.70 per cent, respectively. Thus, this period can be validly interpreted as one of relative monetary tightening in Germany, the main component being the rise in the reserve ratio on July 1 and the other being the fall in external interest rates relative to German interest rates.

In view of this policy of monetary tightening, it is not surprising that the previous rate of capital inflow was maintained in July, with a recorded short-term capital inflow of DM 1.57 billion and a balance of unclassifiable transactions of DM 1.79 billion. In August, with no further measures of monetary tightening, recorded short-term capital inflow was DM 1.03 billion and the balance of unclassifiable items was only DM 0.14 billion, total short-term inflow in August amounting to one third of the rate in July.

On September 1, 1970 effective marginal reserve ratios were raised substantially,28 except for the abolition of the 30 per cent requirement pertaining to liabilities to nonresidents. The net effect of the reserve ratio increases was estimated to drain DM 3.5 billion of liquidity from the banks—i.e., to require an additional 3.5 billion of assets in the form of reserves.29 Once again, the response to this additional degree of monetary tightness turned out to be substantial capital inflow, with recorded short-term capital inflow plus the balance of unclassifiable transactions amounting to DM 3.3 billion in September with DM 1.0 billion coming through the German banks. In other words, from a statistical point of view, the monetary policy measures were almost completely negated by the subsequent capital flows. No additional measures were imposed in October, when total short-term inflow amounted to only DM 0.5 billion.

In the months of November and December 1970, reductions were made in the discount rate to 6.5 per cent and 6 per cent, respectively.30 However, yield differentials remained strongly in favor of Germany, with a substantial capital inflow of DM 2.3 billion plus a rise in the balance of unclassifiable transactions of DM 3.4 billion. The apparent cause was the sharp drop in Euro-dollar rates, with the fall in German rates being significantly less, leading to a substantial average uncovered (three-month) differential in November of 1.63 per cent in favor of Germany, the average covered differential being 1.38 per cent as the forward discount was only 0.25 per cent (see Table 9). In fact, with minor exceptions both uncovered and covered differentials had been strongly in favor of Germany since the increase of 1.5 per cent in the discount rate in March 1970.

Table 9.

Monthly Average of Three-Month Money Market Rates in Germany, 1969–71

(In per Deutschecent per annum)

article image
Source: Deutsche Bundesbank, Monthly Reports.
Result of tight monetary policy in 1970

The overall picture is clearly one of substantial capital inflow resulting from the combined effects of the discount rate and reserve ratio policies pursued by the authorities in the face of declining yields in external markets. Over the year 1970 the sum totals of short-term capital inflows and unclassifiable inflows were DM 14.0 billion and DM 9.7 billion, respectively, with foreign exchange and gold holdings of the Bundesbank increasing by DM 22.7 billion. The money supply expanded more rapidly over this period than it had for the same period in the preceding two years. Inflationary trends persisted, and incomes appeared to be rising at very high rates. The only clear result of the policies was to cause substantial accumulation of foreign exchange reserves.

The events in 1971 to May 31

In the first quarter of 1971 the German three-month money market rate, on a covered basis, averaged 1 per cent higher than the three-month Euro-dollar rate. In April 1971, German interest rates fell sharply, as the result of the reduction in the discount rate from 6 per cent to 5 per cent on April 1, and this removed the covered differential in favor of Germany. In January 1971 there was the usual seasonal pattern of capital flows, i.e., a substantial recorded short-term capital outflow and a substantial net inflow recorded in the errors and omissions component of the balance of payments.31 The sum total of the net short-term inflows and errors and omissions over the five months prior to the end of May 1971 was DM 17.7 billion.32 To this sum total could be added the rather exceptional net long-term inflow of DM 1.3 billion in April 1971, reflecting large long-term advances and loans from foreign sources.

On May 9, 1971 Germany allowed the exchange value of the deutsche mark to float outside the previously established 1 per cent margins. On June 1, the Bundesbank increased minimum reserve requirements by over DM 5 billion in an effort to neutralize the liquidity effects of the speculative capital inflows. In addition, new foreign exchange regulations were introduced to discourage certain types of borrowing from nonresidents.

Les flux de capitaux peuvent-ils mettre en échec la politique monétaire ? L’expérience allemande


Cette étude a pour objet d’établir le rapport existant entre les décisions de politique monétaire prises en Allemagne de 1963 à 1970 et les flux de capitaux et en particulier, d’examiner si le recours aux modifications des coefficients de réserves obligatoires est un moyen efficace de neutraliser les effets de flux de capitaux sur la masse monétaire. Fait important : la manipulation des coefficients de réserves moyennes obligatoires – point essentiel de contrôle en matière de politique monétaire allemande - est rapidement et sérieusement contrebalancée dans son incidence sur les liquidités du système bancaire par des entrées considérables de capitaux qui, pour l’essentiel, sont repris au poste “Erreurs et omissions” de la balance des paiements. Cette observation semble étayer empiriquement la théorie selon laquelle la marge de manoeuvre dont dispose en politique monétaire un pays, comme l’Allemagne, qui permet la libre circulation de capitaux financiers à travers ses frontières, est fort limitée en régime de taux de change fixes.

L’auteur utilise des données mensuelles pour la période 1963 - 70 et le modèle d’ajustement de portefeuille a été modifié, retenant comme variables indépendantes-clefs : a) les changements décalés ou non sur trois mois des rendements différentiels couverts en Allemagne et dans le compartiment euro-dollars; b) les modifications de la réserve obligatoire. Le résultat le plus important enregistré c’est que le relèvement des coefficients de réserves obligatoires, qui se produit toujours en Allemagne au début du mois, est neutralisé vers la fin du mois à concurrence de 61 à 86 pour cent, la plupart des flux étant enregistrés au poste “Erreurs et omissions” dans la balance des paiements. Ce résultat semble indiquer que les emprunts contractés à l’étranger par les sociétés ont rendu inopérantes, dans une large mesure, les restrictions de la politique monétaire allemande, comme ce fut le cas, par exemple, en 1970-71 (point développé à l’Appendice de l’étude).

Les changements des différentiels sur les rendements couverts à terme en Allemagne et dans le marché de l’euro-dollar sont jugés assez caractéristiques comme variable explicative dans les équations capital-flux. Il y a aussi clairement décalage de réponse, 60 pour cent ayant lieu dans les deux mois et 73 pour cent dans les trois mois. Une réduction de 1 pour cent du taux sur l’euro-dollar, les taux pratiqués en Allemagne restant inchangés, semble entraîner dans ce pays des entrées de capitaux de l’ordre de 5,3 milliards de DM sur une période de six mois.

Il est démontré que, par le biais des équations simultanées, les résultats sont faussées de sorte que le différentiel couvert en faveur de l’Allemagne tend à réagir aux flux de capitaux. Dans la mesure où ceux-ci ont un effet stabilisateur, c’est-à-dire que les entrées de capitaux réduisent le différentiel couvert, les distorsions qui apparaissent dans les estimations de l’équation unique reflètent des sous-estimations tant de la mise en échec de la politique monétaire que de la sensibilité des flux de capitaux au taux d’intérêt.

Los flujos de capital como neutralizadores de la política monetaria : La experiencia alemana


En este estudio se examina la relación entre determinadas medidas de política monetaria adoptadas en Alemania durante el período 1963–70 y los flujos de capital que han tenido lugar después, y en particular se somete a prueba la eficacia de recurrir a las modificaciones de los coeficientes de reserva obligatoria como medio de neutralizar el efecto que ejercen los flujos de capital en la oferta monetaria. Un descubrimiento importante es el de que las modificaciones de las reservas obligatorias medias—el principal control de política monetaria en Alemania—quedan rápida y sustancialmente neutralizadas, en cuanto a su efecto sobre la liquidez bancaria, por entradas de capital registradas en la partida “errores u omisiones” de la balanza de pagos. Este resultado proporciona apoyo empírico a la noción de que en un sistema de tipos de cambio fijos son muy limitadas las opciones de política monetaria disponibles en un país como Alemania, que permite el movimiento libre de capital financiero a través de sus fronteras.

En este estudio se utilizan datos mensuales para el período 1963–70 y se adopta un modelo modificado de ajuste de la cartera de activos financieros, para los flujos de capital, en el que las variables clave independientes son las variaciones, desfasadas y no desfasadas, en el margen de diferencia entre el interés alemán con cobertura a tres meses y el del eurodólar, y las variaciones en las reservas obligatorias. El resultado principal es que el aumento de las reservas obligatorias, que se lleva a efecto sólo a principios de mes en Alemania, queda neutralizado entre el 61 por ciento y el 86 por ciento al término del mes, registrándose la mayoría de los flujos en la partida “errores u omisiones” de la balanza de pagos. Este resultado sugiere que los préstamos obtenidos por empresas en el exterior han coartado en gran medida las políticas monetarias restrictivas de Alemania, como ocurrió, p. ej., en 1970–71 (situación que se documenta en el Apéndice).

Se ha encontrado que la variación de los márgenes de diferencia entre los tipos con cobertura alemán y del eurodólar es significativa como variable explicativa en las ecuaciones de flujos de capital. Hay prueba también de una reacción desfasada, ocurriendo el 60 por ciento en dos meses y el 73 por ciento en tres meses. Se ha observado que una disminución del 1 por ciento en el tipo del eurodólar, permaneciendo invariables los tipos alemanes con cobertura, induce una entrada total en Alemania de DM 5.300 millones en un período de seis meses.

Se demuestra que en los resultados hay un sesgo de ecuaciones simultáneas, puesto que el margen de diferencia con cobertura a favor de Alemania tiende a reaccionar ante los flujos de capital. Siendo estabilizadores dichos flujos, ya que los flujos de capital reducen el margen de diferencia con cobertura, se deduce que los sesgos que aparecen en la ecuación sencilla reflejan subestimaciones tanto de la neutralización a la política monetaria como del grado de sensibilidad de los flujos de capital con respecto a los tipos de interés.


Mr. Porter, economist in the Financial Studies Division of the Research Department, is a graduate of the University of Adelaide (Australia) and Stanford University (California). He formerly taught at Simon Fraser University (British Columbia, Canada) and visited the University of Essex (England) on a Canada Council Research Grant in 1969–70. This paper benefited from discussions in seminars at the University of Bologna and Purdue University. Giorgio Basevi, Peter Barton Clark, and Patrick Hendershott are to be thanked for helpful comments.


The monetary events of 1970–71 are discussed in the Appendix.


In the Monthly Report of the Deutsche Bundesbank, June 1971, p. 6, it is evident that the Bundesbank has reached a similar conclusion: “Any reduction in domestic liquidity and any resultant tightening of credit conditions such as would have appeared directly or indirectly as a consequence of credit policy or fiscal policy, would in the last analysis merely have intensified the tendency to attract foreign money. The extent to which in this context it was a case of Sisyphus’ labors is apparent from the (following) figures on the creation and destruction of bank liquidity during the period between the beginning of 1970 and May of this year, when the Deutsche Bundesbank’s obligation to intervene was cancelled.”


A useful reference list may be found in E. E. Learner and R. M. Stern, Quantitative International Economics (Boston, 1970), p. 105. A recent study is by William Branson and Raymond D. Hill, “Capital Movements in the OECD Area: An Econometric Analysis,” OECD Economic Outlook, Occasional Studies, December 1971.


There is a great deal of literature available on this subject. See, for example, the essays in R. A. Mundell, International Economics (New York, 1968), especially Chapters 11, 16, and 18; and R.I. McKinnon and W. R. Oates, “The Implications of International Economic Integration for Monetary, Fiscal and Exchange Rate Policy,” Princeton Studies in International Finance, No. 16, 1966.


A brief discussion of these points may be found in the Report of the Deutsche Bundesbank for the Year 1970, pp. 17–18.


An unpublished paper which derives capital flow equations from a general equilibrium portfolio model, including money demand and supply equations, is by Pentti Kouri and Michael G. Porter, “International Capital Flows and Portfolio Equilibrium,” 1972.


D. Mathieson of Columbia University in an unpublished study using annual data for the period 1961–68 found that 80 per cent of the change in (base) money demand was satisfied through foreign sources.


A useful summary of this approach is given in Branson and Hill, op. cit., pp. 5–11.


Given that there is much greater volatility in interest rates than in the level of wealth on a month-to-month basis, it follows that the fluctuations in capital flows are most likely to be the result of variation in interest rates. In this case the omission of the wealth variable is a less serious problem than it would otherwise be.


For a study which focuses on this problem, see Zoran Hodjera, “Short-Term Capital Movements of the United Kingdom, 1963–67,” Journal of Political Economy (July-August 1971).


Strictly speaking, it would be more nearly correct to measure Lt at the beginning of the month, rather than to use the published average monthly data; however, the effect of making this correction is extremely small.


It should be noted that n is the average of the reserve ratios applying to German banks, since the different classes of banks are subject to different required reserve ratios. The definition of liabilities includes liabilities to nonresidents and hence the change in required reserves reflects changes in the required ratio with respect to the level of, and growth in, liabilities to nonresidents. It should also be noted that in its own analysis of officially induced changes in bank liquidity the Bundesbank uses the broader concept (ARR and not ARR*).


The author is indebted to Pentti Kouri of the Research Department for suggesting this approach.


Branson and Hill, op. cit., pp. 34–35.


There is a well-known tendency to “window-dress” balance sheets at the end of the calendar year.


For example, prior to the 1969 revaluation, the Bundesbank offered inducements to German banks to place funds abroad, the inducement being in the form of a preferential rate for the forward purchase of deutsche mark.


Branson and Hill, op. cit.


The Almon lag technique is used. See Shirley Almon, “The Distributed Lag Between Capital Appropriations and Expenditures,” Econometrica, Vol. 33 (January 1965), pp. 178–96. The polynomial was unconstrained for reasons set out by P. J. Dhrymes, Distributed Lags: Problems of Estimation and Formulation (San Francisco: Holden-Day, 1971), p. 232. The choice of the second-order polynomial reflected the expectation that the pattern of response might well be one of either quadratically declining weights or of an inverted V formation, both possibilities being allowed in an unconstrained second-order polynomial.


Manfred Willms, “Controlling Money in an Open Economy: The German Case,” Federal Reserve Bank of St. Louis Review (April 1971).


The actual estimates of the offset within one month varied from 61 per cent to 86 per cent; however, the presence of simultaneous equations bias means that these are underestimates of the true coefficients.


In the three months prior to April 1971, for example, private German borrowing in foreign countries was at approximately the same level as the entire lending by the German banking system to domestic customers. Thus, half the borrowers were paying the substantially lower Euro-dollar rates.


On March 9, 1970 the Bundesbank raised the discount rate from 6 per cent to 7.5 per cent and the rate for advances against securities from 9 per cent to 9.5 per cent. On April 1, 1970 additions to bank external liabilities over the level on March 6, or the average level on the return dates in February, were subject to a supplementary minimum reserve of 30 per cent.


Monthly Report of the Deutsche Bundesbank, March 1970, p. 5.


Monthly Report of the Deutsche Bundesbank, May 1970, p. 72.


See Victor Argy, “Monetary Policy and Internal-External Balance,” Staff Papers, Vol. XVIII (1971), pp. 508–26, for an analysis of the effect of various reserve requirements on the marginal cost of different sources of bank funds.


Monthly Report of the Deutsche Bundesbank, July 1970, p. 5.


The average reserve ratio of the banking system increased from 6.3 per cent in June to 7.3 per cent in July. The actual measure increased the minimum reserve ratio by 15 per cent over the ratios on July 1, 1970.


From September 1, 1970 additional reserves were required to be held on the growth of liabilities, the ratio being 40 per cent for sight and time liabilities and 20 per cent for savings deposits. Growth was to be assessed relative to the average level in the second quarter of 1970.


Monthly Report of the Deutsche Bundesbank, September 1970, p. 8.


An additional measure introduced December 1, 1970 was the restoration of the 30 per cent reserve requirement on the growth of foreign liabilities.


See Tables 1–5 for estimates of the significance of seasonal factors in December and January.


The May 1971 figures are provisional.